Enterprise Agile Rollout at AOL

Experience Report Published on PMI Agile Community of Practice Newsletter February 2012: Enterprise Agile Rollout at AOL

This report is a reflection of how AOL identified a need to change development methodologies and grow team creativity and ownership.  The company reinvested in agile training and support over the past twelve months with the goal of a complete paradigm shift for the technology arm.  This report below points out some critical decisions, adaptations, and the responsiveness to change during the rollout of Agile through training and coaching.

Background

An effort had been made in recent years to move to Agile but did not consistently receive the support necessary to embrace and sustain it.  In late 2010, there were two major agile tools in use globally across AOL.  A consolidation of agile tools was made in late 2010 to cut costs and to begin to centralize the company backlog.    Many teams had been trained in Scrum over the years so during the early planning stages of the transition, it was assumed the tool switch would be simple and quick.  Tool training sessions were held and it was during this time we discovered how many teams still needed basic agile concepts, tool use, team alignment and training.

In February of 2011, I took on a Program Director role to evaluate training needs, manage the training vendor relationship, implement and manage our corporate agile tooling system, build an internal agile community of agile experts, monitor progress and make recommendations to the CTO and his technical cabinet.  I managed this function with the addition of two people on my team, both with strong project management experience but neither considered themselves to be agile experts when they started on the program.  With this small team, we took on three major challenges. First, we needed to figure out how many teams needed training whether they knew it or not (hint hint). Second, we had to figure out how to handle the requests for tool training for teams that did not have a clear understanding of what agile really was.  And lastly, we needed to do all of this as cost effectively as possible so we could reach over 1,500 people across Technologies and our business organization within our approved and assigned budget.

We made a few assumptions in the beginning.  Training would be team-based so that we weren’t simply teaching people about a new set of techniques and tools but rather helping teams to start thinking and behaving in agile ways and provide a baseline to begin their journey as a team that we would revisit at key intervals to monitor their progress.  We selected Scrum and Kanban as the two standard styles of agile courses.  Teams that were not already in our standard agile tool would wait until after they completed their agile training before we would train them in the tool.  This rollout would be tracked through a standard agility survey at three different points in our rollout: a baseline in March, a six-month check and a twelve-month final survey. My team’s goal was to ensure that the rollout was not simply “complete” after we trained everyone; we needed to find a way to sustain the investment past the classroom and continue to see improvements.

Please do visit the PMI Agile Community of Practice site http://agile.vc.pmi.org to raise questions and discuss this report.

Getting Organized

To begin, we created a roster of all the teams in the company and began estimating how many Scrum and Kanban classes we’d have to schedule. This was a challenge that even HR could not have solved for us.

 When we talk about “Teams” in agile it’s usually related to either a Scrum team of 5-9 people or a Kanban team of up to 35 people.  These teams should be set up independent of organizational lines so that they are fully functional teams and can remain together despite changes in reporting structure.  However, as we began to pull our rosters together, we would get Scrum Team rosters of over 40 people or sometimes only 2 people.  Based on this discovery, we spent time working with managers, development teams, business leaders and product owners to help them understand the difference between an agile team and an org chart team.  This helped us define the correct team sizing, establish a management view of how their teams would be working together during the training as well as after the training and determine the number of classes that would be necessary to consider our classroom training phase complete.

Setting a Standard

Having already started down the agile path over the years, the popularity of Scrum still had a strong influence.  However, Kanban was getting some interest in one of our larger consumer product areas so we decided to offer training in either Scrum or Kanban as our two standard agile practices.  We made sure to educate people about the differences, the pros and cons before teams made their decision on which practice to use.

Figure 1: What AOL thought they were doing

While there are many different thoughts on how to choose between the two styles, we tried our best to stay agnostic and allow the teams to make the choice.  In most cases, teams that felt they had a mature and seasoned group of teammates and needed little to no guidance chose Kanban.  Teams that chose Scrum based it either on prior knowledge or their opinion that heavily date driven work would work best in Scrum to keep them focused on delivery dates.  Understand that while some agilists would try very hard to dispel myths and correct false assumptions spending valuable cycles debating with teams, we felt it would support our efforts to drive the self-management culture by allowing teams to make these decisions for themselves rather than dictate them or have management decide for them.

I’m pleased to share that overall, this was successful even when some teams stumbled along following their own path.  The Classic Mail and Phoenix Mail teams started with Kanban and after about three months choose the go back to Scrum.  We also learned to be a more flexible rather than push the purist approach to Scrum and Kanban.  Many of us in the agile community with strong opinions and valid experience have a difficult time with this lax attitude but it was a lesson learned that when working with such a large globally operated company, there is no cookie cutter way to gain trust and adoption from the very teams you work with.

We began using the phrase agile with a small “a” rather than agile with a capital “a” so that people would understand we were trying to adjust our approach and recognized personalization of a team’s agile practice was indeed very personal and should be honored during conversations, coaching and formal training courses and we continue to support this approach today.

Agile Tool Solution

Several teams were eager to jump into our standard agile tool even though they were not practicing Scrum or Kanban specifically. For example, some teams believed they were a Scrum team but their sprints lasted months instead of weeks, their sprints weren’t all the same length and they often had a single team working on more than one sprint at a time.  All of these were red flags to investigate further into what the teams were doing, why the made those decisions, confirm some of their terminology, and determine the next step to help them, not to bulldoze in and “correct” them.   Despite our advice, some teams plowed forward with the tool only to become frustrated once they realized what they were practicing may have been agile but wasn’t necessarily Scrum and was difficult to adopt our tool. There were some teams that slowly abandoned the tool all together or left the project managers to do all the entries and monitoring in the tool.

I must note here that the tool we used was heavily Scrum bound so for Kanban teams, it required a bit of customization and workarounds which, for a new agile team, was a big distraction from focusing on getting work done and learning a new way of working together. To minimize this, each team was encouraged to continue using the physical board from their training (Scrum Story Boards and Kanban Boards) and move it into their work area for a few cycles or sprints.  While the team used the boards to reflect their work in progress and plans, the Scrum Master, Kanban Leads or Project Managers would set up the tool mimicking the progress in the tool.  This allowed some time for the leads to become proficient in the tool, make adjustments to the set up and allow the team to gain confidence in their agile practice before launching the full team into the tool.  Teams that followed that path, seem to have had less frustration with their agile practice and they recommend it to other new teams.

Agility Surveys

Fortunately, our rollout was not given a hard deadline but we did have a set budget that we estimated would last about 10-12 months.  Knowing that large change efforts of any kind take time, momentum, adoption and support; we created an online survey based on a template provided by our training vendor and gathered a baseline response from everyone in the company willing to offer their input.  The agility survey contained questions to determine each person’s perspective about how things were going in these categories: Product Ownership, Release Planning & Tracking, Testing Practices, Planning & Tracking, Development and the Agile Team.

The results of the baseline from March 2011 were shared with the agile trainers to provide awareness on potential training opportunities and the leadership team of each organization to reinforce the need for the investment in training and coaching.  In November 2011, teams took the survey once again and the results showed between 8% and 14% improvement over the baseline.  It was useful to see where people believed they were taking a step backwards and where they were leaping forward.

We presented the survey as

• a pulse of how people believe their teams are doing
• a snapshot of changes over time
• an indicator of positive and negative impacts
•  only a sample of people willing to provide their feedback

and most importantly

• a conversation starter.

The survey was NOT

• equal to product success
• enough information on its own
• finished

We still need to dig deeper through conversations around these results.  While it does not always equal revenue and product success; it can help.  We accepted that it would show purely the perception that our teams had on where they were at a point in time during the rollout.  We also hoped we could supplement further evaluations such as coaching observation, Product and team level metrics to make correlations between them and the agility survey.  Our next survey will be in March 2012 and we look forward to seeing steady improvements.

Ensuring Long Term Sustainment

It was apparent very early in our rollout that a three person rollout team was not enough to support a large scale rollout.  We were not so ignorant to think we could do it all alone but we were wise enough to realize that we would slowly regress from all the work we did in the classroom if we didn’t invest in creating leadership from within our own people.

Because we were so dedicated to spending our money efficiently and wisely, rather than hire more people on our rollout team, we asked for the senior leaders in each organization to nominate an individual for the “Agile Champion” role.  This person would be the central point of contact for their teams to funnel training requests, answer basic questions about the rollout and even offer coaching if they were comfortable.  They would liaise with the rollout team each month in an Agile Champion meeting where the status of training requests, survey results and Q&A would be covered. Over time, these champions provided not only a buffer and filter of the overabundance of questions and training requests but they also became our external and internal champions, supporting us in places where we could not be.

Over the summer of 2011, nine teams from HR/Legal, Advertising, Identity Services, AOL Mail and AOL Search had taken either basic Scrum or Kanban courses but they struggled alone without access to experts in the days after class and we began to see the need of post-class onsite coaching to help when needed.  This lead to our monthly Agile Leader meetings where we invited anyone from around the company to join us for an hour to discuss any topic, ask any question and even do a few show-n-tells.  Several people started to really shine in these sessions.  They became our next generation of agile trainers that we use today and have been very successful because not only are they experienced having gone through training with their own teams but they WANT to help people in other teams even if it’s not in their job description or their organization.   To have passionate, experienced and engaging people working towards the same goals made the rollout the success it continues to be today.  This is where we really started to make a difference in sustainment.

Scaling an Agile Rollout

At the beginning of the rollout, we attempted to get each team through the training courses as quickly as possible.  We started in early February with 12 teams on the West and East Coasts in the midst of launching Scrum and Kanban and the list of teams waiting for available trainers quickly grew faster than we could deliver.  We also noted that after teams completed their training, they often needed help but our trainers were already booked to begin training the next team.  It was becoming increasingly difficult to give the time and attention to the teams that had completed training and they often struggled alone and frustrated with no one to help them apply all of their new training and skills.

This is about the time we began to reevaluate our rollout model and worked with our training vendor to make necessary adjustments.  Rather than plow through the training requests just to get every team through the training, we selected a set of product areas and teams that showed the most interest, had high visibility projects and the best executive support to focus our training dollars and devoted our attention to them for 3 months with classroom training, and leadership coaching. MapQuest, AOL Classic Mail, AOL Phoenix Mail, Anti-Spam, and AOL Search made up these core focus teams, which we called our “Wave 1” group. By providing these teams with a dedicated agile coach to help them through their adoption post-classroom, they were able to give attention to their practice and their tool adoption with help readily available without significant delays that we experienced early in the rollout.

With this shift in our attention and training dollars, a waitlist of training requests was created.  Because of this waitlist, some of our agile volunteers were eager to learn more and volunteered to help train the teams that were not included in the Wave 1 group.  They attended Wave 1 team training sessions, learned how the trainers worked through the materials and assisted in writing a mini-training course that could be offered to those teams not in the Wave 1 group.

By November of 2011, we had a library of course materials including video training that our volunteer trainers created.  The library included helpful guides, links to sites and recommended books, videos of Scrum, Kanban and even Scrumban mini-courses that our volunteer trainers had completed.  They were able to successfully provide a training alternative to waitlisted teams that did not want to delay their own agile adoption.  With these course materials and resources available to anyone at the company, the need for additional training funds lessened and more teams were already adopting the culture of helping each other and becoming self-managing by doing the research themselves and trying new things on their own.

By the end of 2011 we had helped over 114 teams from the US, UK, Ireland, Germany, Israel, and India.  We had launched new agile teams, help agile teams tune up their practices, assisted teams in adopting our agile tool and coached teams through particularly hard situations when they needed an unbiased third party to intervene.  We were able to do all of this because of several key volunteers who stepped up to keep the momentum of agile adoption going.  I’m extremely proud of such a small group of people that took the time to learn more, share what they learned and continue to offer their time to other people.  We have created a network of leaders, coaches and trainers that we would not have without their contributions.

Summary

In my experience, a balance between outside expertise and internal champions is important to increase the return on the agile investment.  It is tempting for us to believe agile is a quick and simple approach to get fast results from our teams when we begin adopting agile into our organization.  It is also common that we trustingly usher a crew of agile consultants into their “work homes” to take control and whip everyone in to shape through a few classes.  Hiring knowledgeable, quality people with practical agile experience can be a challenge when our own understanding is limited and we are unaware of what it takes to undertake such an enormous task of changing a culture, changing behaviors, and putting the right people together.

In order to make your consulting dollars really pay off, invest in spreading the knowledge and expertise within your own people so that the learning keeps going and the energy to continue is not dependent on your budget.  But remember that the learning should not stop and while your teams may have completed their initial training, access to a knowledgeable coach can make the difference between success and frustration, minor improvements and steady improvements and agile buzzwords and agile behaviors.  One of the indicators I look for in a great agile team is when they are not satisfied with the status quo and they continuously look for ways to improve without being prompted to do so.  Those are the teams to showcase and encourage.  As the teams share, they become more confident and as more teams see what success looks like, they more they will be open to adopting those new ways of working together.

About the Author

Christen is an Agile Evangelist at AOL and is currently responsible for agile training and support across the company.  She is also program managing a large technology initiative to centralize all user identity for AOL products such as AOL Mail, AIM and AOL Music.  She provides coaching and mentoring over 114 different AOL Technology teams around the globe and ensures that all training needs are met.  Christen has over eighteen years of experience in a variety of roles and environments and is passionate about using Agile methods to help guide teams in the implementation of quality products and deliverables that would stand out as successful examples to other teams and encourage Agile practices at an organizational level.  She was so convinced after getting her CSM in 2006 that agile was that “special sauce” for building great teams that she voluntarily began offering a simple Scrum class to anyone in the company that wanted to learn.  The changes caught the attention of the Technology leaders, at which point she was asked to integrate agile techniques and practices into the traditional waterfall SDLC model for AOL.  This transition led to their interest in full adoption of agile in 2007 and a large re-investment in 2011.

Christen is a requested speaker at local events in the DC area and an active blogger and contributor in the agile community. Christen is a certified Project Management Professional and a Scrum Alliance Certified Scrum Professional.